Case study · Etnia Eyewear Culture

AI chatbot to query your data in natural language

You ask in Teams. Answers in seconds. Without writing a single query.

Agile, express development

Cloud-native, scalable architecture

User-centric approach

Where it all started

The data was there,
the answers were not

Etnia Eyewear Culture has a data architecture that matches the complexity of its business. The data was there. Getting answers out of it was the hard part.

Its analysts needed to go beyond the BI dashboards. Specific, day-to-day questions. And for each one, the same route: design a SQL query, validate it, run it. About 20 minutes. Per query.

Multiply that by every question, every team, every day.

The challenge

Specific questions the BI dashboards did not answer.

A different SQL query for each request: design, validate, run.

About 20 minutes on average per question, before any fine-tuning.

And always a technical profile in the middle.

What we proposed

A chatbot in Microsoft Teams that answers questions about customers, sales, stock and tickets. From a single chat. In natural language.

The shift

Day and night

All it took was letting anyone talk to their data. The rest, the numbers tell it.

Before

A SQL query for every question

About 20 minutes of waiting per query

You depend on the technical team

Inconsistent results between queries

Now

A question in natural language

An answer in seconds

You solve it yourself

Structured, consistent information

less response time per query

0 %

returned to each user, every month

0 h

more data queries than before implementation

0 %

The value generated

Beyond the numbers

Data within everyone's reach

Precise answers in natural language, with no need to go through the technical team.

Faster decisions

Five business areas deciding on data in seconds. With no intermediaries.

No technical barriers

Questions in Microsoft Teams. Zero dashboards you have to learn to navigate.

Ready to grow

One channel today, as many as needed tomorrow. The architecture holds with no changes.

More trust in the data

Instant answers straight from the source. The data gets used more because it is trusted more.

Scale without adding headcount

Unlimited queries at once. With no need to hire anyone or overload the teams.

What the client says

We used to depend on manual queries and on waiting for someone to validate or run a query. Now anyone gets information on the spot, with no technical knowledge. We have gone from hours or days to seconds.

BI Data Specialist

Etnia Eyewear Culture

F&Q

Frequently asked questions

What is a conversational SQL chatbot?

It is an assistant that turns your natural-language question into a SQL query, runs it against your database and returns the answer in seconds. On the inside, an AI agent interprets what you ask, knows the structure of your tables and builds the right query. On the outside, you just write as you would to a colleague. What used to require a technical profile and a custom query is now solved in a sentence. And because the answer comes from your own data, it is precise and the same every time you ask.

How is the chatbot different from a BI dashboard like Power BI?

Your BI is perfect for what you already know you are going to look at: the usual KPIs, updated and well presented. The challenge comes with new questions, the ones no one anticipated when the report was designed. That is when you have to wait for someone to create a view or write a query. The chatbot covers exactly that gap: the long tail of specific day-to-day questions. It works alongside your BI: the BI handles the predictable; the chatbot, whatever comes up on the fly. Together they cover the two halves of analysis.

Are the answers reliable, or can it hallucinate?

That is the right question. The short answer: it does not make them up. The AI provides the language; your system provides the data. Instead of guessing, it generates a SQL query that runs against your real database and returns what is actually in it. That is why the answer is consistent (ask the same thing, get the same thing) and traceable: every result can be checked against the source. That is the difference between an assistant you chat with and one you can base a decision on.

Is it secure, and who can see each piece of data?

Yes. Control is role-based: each person queries only the information that corresponds to them, relying on the permissions that already exist in your organization. Everything runs on Microsoft Azure and your data stays within your environment, without going out to third parties. For sectors with regulatory demands, that governance of access matters as much as the answer itself: knowing who asks what is part of the value.

What does it take to implement it in my company?

At its base, a clean, consolidated data source (a SQL Server or a data warehouse) and Microsoft Teams, where your team already works. From there we move in phases: we start with a defined scope (one area, a few frequent questions), get it running and train it by adding real use cases. That way you see value early and the system grows at the pace of your business. It is the same agile approach with which we took this project from a first version to a daily-use tool.